In November 2024, John Kim developed a new evaluation metrics suite for NLP models in the kilian-group/phantom-wiki repository, focusing on precision, recall, and F1 scoring functions. Using Python and leveraging his data science and machine learning expertise, he designed these metrics to provide nuanced model performance assessment and support data-driven tuning. The implementation established a clear API for future metric extensions, laying the groundwork for scalable evaluation pipelines. By maintaining clean, traceable changes in a focused commit, John ensured reproducibility and maintainability. His work addressed the need for robust benchmarking, enabling more informed deployment and ongoing performance tracking.